深度学习网络模型————Swin-Transformer详细讲解与代码实现
时间：2024-04-01 12:05:34 来源：网络cs 作者：璐璐 栏目：卖家故事 阅读：
阅读本书更多章节>>>>
论文名称：Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
原论文地址： https://arxiv.org/abs/2103.14030
官方开源代码地址：https://github.com/microsoft/Swin-Transformer
深度学习网络模型——Swin-Transformer详细讲解与代码实现
一、网路模型整体架构二、Patch Partition模块详解三、Patch Merging模块四、W-MSA详解五、SW-MSA详解masked MSA详解 六、 Relative Position Bias详解七、模型详细配置参数八、重要模块代码实现：1、Patch Partition代码模块：2、Patch Merging代码模块：3、mask掩码生成代码模块：4、stage堆叠部分代码：5、SW-MSA或者W-MSA模块代码： 九：模型整体流程代码实现：论文名称：Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
原论文地址： https://arxiv.org/abs/2103.14030
官方开源代码地址：https://github.com/microsoft/Swin-Transformer
一、网路模型整体架构
二、Patch Partition模块详解
三、Patch Merging模块
四、W-MSA详解
五、SW-MSA详解
masked MSA详解
六、 Relative Position Bias详解
七、模型详细配置参数
八、重要模块代码实现：
1、Patch Partition代码模块：
class PatchEmbed(nn.Module): """ 2D Image to Patch Embedding split image into non-overlapping patches 即将图片划分成一个个没有重叠的patch """ def __init__(self, patch_size=4, in_c=3, embed_dim=96, norm_layer=None): super().__init__() patch_size = (patch_size, patch_size) self.patch_size = patch_size self.in_chans = in_c self.embed_dim = embed_dim self.proj = nn.Conv2d(in_c, embed_dim, kernel_size=patch_size, stride=patch_size) self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity() def forward(self, x): _, _, H, W = x.shape # padding # 如果输入图片的H，W不是patch_size的整数倍，需要进行padding pad_input = (H % self.patch_size[0] != 0) or (W % self.patch_size[1] != 0) if pad_input: # to pad the last 3 dimensions, # (W_left, W_right, H_top,H_bottom, C_front, C_back) x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1], # 表示宽度方向右侧填充数 0, self.patch_size[0] - H % self.patch_size[0], # 表示高度方向底部填充数 0, 0)) # 下采样patch_size倍 x = self.proj(x) _, _, H, W = x.shape # flatten: [B, C, H, W] -> [B, C, HW] # transpose: [B, C, HW] -> [B, HW, C] x = x.flatten(2).transpose(1, 2) x = self.norm(x) return x, H, W2、Patch Merging代码模块：
class PatchMerging(nn.Module): r""" Patch Merging Layer. 步长为2，间隔采样 Args: dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, dim, norm_layer=nn.LayerNorm): super().__init__() self.dim = dim self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) self.norm = norm_layer(4 * dim) def forward(self, x, H, W): """ x: B, H*W, C 即输入x的通道排列顺序 """ B, L, C = x.shape assert L == H * W, "input feature has wrong size" x = x.view(B, H, W, C) # padding # 如果输入feature map的H，W不是2的整数倍，需要进行padding pad_input = (H % 2 == 1) or (W % 2 == 1) if pad_input: # to pad the last 3 dimensions, starting from the last dimension and moving forward. # (C_front, C_back, W_left, W_right, H_top, H_bottom) # 注意这里的Tensor通道是[B, H, W, C]，所以会和官方文档有些不同 x = F.pad(x, (0, 0, 0, W % 2, 0, H % 2)) # 以2为间隔进行采样 x0 = x[:, 0::2, 0::2, :] # [B, H/2, W/2, C] x1 = x[:, 1::2, 0::2, :] # [B, H/2, W/2, C] x2 = x[:, 0::2, 1::2, :] # [B, H/2, W/2, C] x3 = x[:, 1::2, 1::2, :] # [B, H/2, W/2, C] x = torch.cat([x0, x1, x2, x3], -1) # ————————> [B, H/2, W/2, 4*C] 在channael维度上进行拼接 x = x.view(B, -1, 4 * C) # [B, H/2*W/2, 4*C] x = self.norm(x) x = self.reduction(x) # [B, H/2*W/2, 2*C] return x3、mask掩码生成代码模块：
def create_mask(self, x, H, W): # calculate attention mask for SW-MSA # 保证Hp和Wp是window_size的整数倍 Hp = int(np.ceil(H / self.window_size)) * self.window_size Wp = int(np.ceil(W / self.window_size)) * self.window_size # 拥有和feature map一样的通道排列顺序，方便后续window_partition img_mask = torch.zeros((1, Hp, Wp, 1), device=x.device) # [1, Hp, Wp, 1] h_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) w_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) cnt = 0 for h in h_slices: for w in w_slices: img_mask[:, h, w, :] = cnt cnt += 1 # 将img_mask划分成一个一个窗口 mask_windows = window_partition(img_mask, self.window_size) # [nW, Mh, Mw, 1] # 输出的是按照指定的window_size划分成一个一个窗口的数据 mask_windows = mask_windows.view(-1, self.window_size * self.window_size) # [nW, Mh*Mw] attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) # [nW, 1, Mh*Mw] - [nW, Mh*Mw, 1] 使用了广播机制 # [nW, Mh*Mw, Mh*Mw] # 因为需要求得的是自身注意力机制，所以，所以相同的区域使用0表示，；不同的区域不等于0，填入-100，这样，在求得 attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) # 即对于不等于0的位置，赋值为-100；否则为0 return attn_mask4、stage堆叠部分代码：
class BasicLayer(nn.Module): """ A basic Swin Transformer layer for one stage. Args: dim (int): Number of input channels. depth (int): Number of blocks. num_heads (int): Number of attention heads. window_size (int): Local window size. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True drop (float, optional): Dropout rate. Default: 0.0 attn_drop (float, optional): Attention dropout rate. Default: 0.0 drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. """ def __init__(self, dim, depth, num_heads, window_size, mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False): super().__init__() self.dim = dim self.depth = depth self.window_size = window_size self.use_checkpoint = use_checkpoint self.shift_size = window_size // 2 # 表示向右和向下偏移的窗口大小 即窗口大小除以2，然后向下取整 # build blocks self.blocks = nn.ModuleList([ SwinTransformerBlock( dim=dim, num_heads=num_heads, window_size=window_size, shift_size=0 if (i % 2 == 0) else self.shift_size, # 通过判断shift_size是否等于0，来决定是使用W-MSA与SW-MSA mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, drop=drop, attn_drop=attn_drop, drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path, norm_layer=norm_layer) for i in range(depth)]) # patch merging layer 即：PatchMerging类 if downsample is not None: self.downsample = downsample(dim=dim, norm_layer=norm_layer) else: self.downsample = None def create_mask(self, x, H, W): # calculate attention mask for SW-MSA # 保证Hp和Wp是window_size的整数倍 Hp = int(np.ceil(H / self.window_size)) * self.window_size Wp = int(np.ceil(W / self.window_size)) * self.window_size # 拥有和feature map一样的通道排列顺序，方便后续window_partition img_mask = torch.zeros((1, Hp, Wp, 1), device=x.device) # [1, Hp, Wp, 1] h_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) w_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) cnt = 0 for h in h_slices: for w in w_slices: img_mask[:, h, w, :] = cnt cnt += 1 # 将img_mask划分成一个一个窗口 mask_windows = window_partition(img_mask, self.window_size) # [nW, Mh, Mw, 1] # 输出的是按照指定的window_size划分成一个一个窗口的数据 mask_windows = mask_windows.view(-1, self.window_size * self.window_size) # [nW, Mh*Mw] attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) # [nW, 1, Mh*Mw] - [nW, Mh*Mw, 1] 使用了广播机制 # [nW, Mh*Mw, Mh*Mw] # 因为需要求得的是自身注意力机制，所以，所以相同的区域使用0表示，；不同的区域不等于0，填入-100，这样，在求得 attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) # 即对于不等于0的位置，赋值为-100；否则为0 return attn_mask def forward(self, x, H, W): attn_mask = self.create_mask(x, H, W) # [nW, Mh*Mw, Mh*Mw] # 制作mask蒙版 for blk in self.blocks: blk.H, blk.W = H, W if not torch.jit.is_scripting() and self.use_checkpoint: x = checkpoint.checkpoint(blk, x, attn_mask) else: x = blk(x, attn_mask) if self.downsample is not None: x = self.downsample(x, H, W) H, W = (H + 1) // 2, (W + 1) // 2 return x, H, W5、SW-MSA或者W-MSA模块代码：
class SwinTransformerBlock(nn.Module): r""" Swin Transformer Block. Args: dim (int): Number of input channels. num_heads (int): Number of attention heads. window_size (int): Window size. shift_size (int): Shift size for SW-MSA. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True drop (float, optional): Dropout rate. Default: 0.0 attn_drop (float, optional): Attention dropout rate. Default: 0.0 drop_path (float, optional): Stochastic depth rate. Default: 0.0 act_layer (nn.Module, optional): Activation layer. Default: nn.GELU norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, dim, num_heads, window_size=7, shift_size=0, mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm): super().__init__() self.dim = dim self.num_heads = num_heads self.window_size = window_size self.shift_size = shift_size self.mlp_ratio = mlp_ratio assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" self.norm1 = norm_layer(dim) # 先经过层归一化处理 # WindowAttention即为：SW-MSA或者W-MSA模块 self.attn = WindowAttention( dim, window_size=(self.window_size, self.window_size), num_heads=num_heads, qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop) self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) def forward(self, x, attn_mask): H, W = self.H, self.W B, L, C = x.shape assert L == H * W, "input feature has wrong size" shortcut = x x = self.norm1(x) x = x.view(B, H, W, C) # pad feature maps to multiples of window size # 把feature map给pad到window size的整数倍 pad_l = pad_t = 0 pad_r = (self.window_size - W % self.window_size) % self.window_size pad_b = (self.window_size - H % self.window_size) % self.window_size x = F.pad(x, (0, 0, pad_l, pad_r, pad_t, pad_b)) _, Hp, Wp, _ = x.shape # cyclic shift # 判断是进行SW-MSA或者是W-MSA模块 if self.shift_size > 0: # https://blog.csdn.net/ooooocj/article/details/126046858?ops_request_misc=&request_id=&biz_id=102&utm_term=torch.roll()%E7%94%A8%E6%B3%95&utm_medium=distribute.pc_search_result.none-task-blog-2~all~sobaiduweb~default-0-126046858.142^v73^control,201^v4^add_ask,239^v1^control&spm=1018.2226.3001.4187 shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) #进行数据移动操作 else: shifted_x = x attn_mask = None # partition windows # 将窗口按照window_size的大小进行划分，得到一个个窗口 x_windows = window_partition(shifted_x, self.window_size) # [nW*B, Mh, Mw, C] # 将数据进行展平操作 x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # [nW*B, Mh*Mw, C] # W-MSA/SW-MSA """ # 进行多头自注意力机制操作 """ attn_windows = self.attn(x_windows, mask=attn_mask) # [nW*B, Mh*Mw, C] # merge windows attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) # [nW*B, Mh, Mw, C] # 将多窗口拼接回大的featureMap shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp) # [B, H', W', C] # reverse cyclic shift # 将移位的数据进行还原 if self.shift_size > 0: x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) else: x = shifted_x # 如果进行了padding操作，需要移出掉相应的pad if pad_r > 0 or pad_b > 0: # 把前面pad的数据移除掉 x = x[:, :H, :W, :].contiguous() x = x.view(B, H * W, C) # FFN x = shortcut + self.drop_path(x) x = x + self.drop_path(self.mlp(self.norm2(x))) return x九：模型整体流程代码实现：
""" Swin TransformerA PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - https://arxiv.org/pdf/2103.14030Code/weights from https://github.com/microsoft/Swin-Transformer"""import torchimport torch.nn as nnimport torch.nn.functional as Fimport torch.utils.checkpoint as checkpointimport numpy as npfrom typing import Optionaldef drop_path_f(x, drop_prob: float = 0., training: bool = False): """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). This is the same as the DropConnect impl I created for EfficientNet, etc networks, however, the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper... See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use 'survival rate' as the argument. """ if drop_prob == 0. or not training: return x keep_prob = 1 - drop_prob shape = (x.shape[0],) + (1,) * (x.ndim - 1) # work with diff dim tensors, not just 2D ConvNets random_tensor = keep_prob + torch.rand(shape, dtype=x.dtype, device=x.device) random_tensor.floor_() # binarize output = x.div(keep_prob) * random_tensor return outputclass DropPath(nn.Module): """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). """ def __init__(self, drop_prob=None): super(DropPath, self).__init__() self.drop_prob = drop_prob def forward(self, x): return drop_path_f(x, self.drop_prob, self.training)""" 将窗口按照window_size的大小进行划分，得到一个个窗口"""def window_partition(x, window_size: int): """ 将feature map按照window_size划分成一个个没有重叠的window Args: x: (B, H, W, C) window_size (int): window size(M) Returns: windows: (num_windows*B, window_size, window_size, C) """ B, H, W, C = x.shape x = x.view(B, H // window_size, window_size, W // window_size, window_size, C) # permute: [B, H//Mh, Mh, W//Mw, Mw, C] -> [B, H//Mh, W//Mh, Mw, Mw, C] # view: [B, H//Mh, W//Mw, Mh, Mw, C] -> [B*num_windows, Mh, Mw, C] windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C) # 输出的是按照指定的window_size划分成一个一个窗口的数据 return windowsdef window_reverse(windows, window_size: int, H: int, W: int): """ 将一个个window还原成一个feature map Args: windows: (num_windows*B, window_size, window_size, C) window_size (int): Window size(M) H (int): Height of image W (int): Width of image Returns: x: (B, H, W, C) """ B = int(windows.shape[0] / (H * W / window_size / window_size)) # view: [B*num_windows, Mh, Mw, C] -> [B, H//Mh, W//Mw, Mh, Mw, C] x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1) # permute: [B, H//Mh, W//Mw, Mh, Mw, C] -> [B, H//Mh, Mh, W//Mw, Mw, C] # view: [B, H//Mh, Mh, W//Mw, Mw, C] -> [B, H, W, C] x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1) return xclass PatchEmbed(nn.Module): """ 2D Image to Patch Embedding split image into non-overlapping patches 即将图片划分成一个个没有重叠的patch """ def __init__(self, patch_size=4, in_c=3, embed_dim=96, norm_layer=None): super().__init__() patch_size = (patch_size, patch_size) self.patch_size = patch_size self.in_chans = in_c self.embed_dim = embed_dim self.proj = nn.Conv2d(in_c, embed_dim, kernel_size=patch_size, stride=patch_size) self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity() def forward(self, x): _, _, H, W = x.shape # padding # 如果输入图片的H，W不是patch_size的整数倍，需要进行padding pad_input = (H % self.patch_size[0] != 0) or (W % self.patch_size[1] != 0) if pad_input: # to pad the last 3 dimensions, # (W_left, W_right, H_top,H_bottom, C_front, C_back) x = F.pad(x, (0, self.patch_size[1] - W % self.patch_size[1], # 表示宽度方向右侧填充数 0, self.patch_size[0] - H % self.patch_size[0], # 表示高度方向底部填充数 0, 0)) # 下采样patch_size倍 x = self.proj(x) _, _, H, W = x.shape # flatten: [B, C, H, W] -> [B, C, HW] # transpose: [B, C, HW] -> [B, HW, C] x = x.flatten(2).transpose(1, 2) x = self.norm(x) return x, H, Wclass PatchMerging(nn.Module): r""" Patch Merging Layer. 步长为2，间隔采样 Args: dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, dim, norm_layer=nn.LayerNorm): super().__init__() self.dim = dim self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) self.norm = norm_layer(4 * dim) def forward(self, x, H, W): """ x: B, H*W, C 即输入x的通道排列顺序 """ B, L, C = x.shape assert L == H * W, "input feature has wrong size" x = x.view(B, H, W, C) # padding # 如果输入feature map的H，W不是2的整数倍，需要进行padding pad_input = (H % 2 == 1) or (W % 2 == 1) if pad_input: # to pad the last 3 dimensions, starting from the last dimension and moving forward. # (C_front, C_back, W_left, W_right, H_top, H_bottom) # 注意这里的Tensor通道是[B, H, W, C]，所以会和官方文档有些不同 x = F.pad(x, (0, 0, 0, W % 2, 0, H % 2)) # 以2为间隔进行采样 x0 = x[:, 0::2, 0::2, :] # [B, H/2, W/2, C] x1 = x[:, 1::2, 0::2, :] # [B, H/2, W/2, C] x2 = x[:, 0::2, 1::2, :] # [B, H/2, W/2, C] x3 = x[:, 1::2, 1::2, :] # [B, H/2, W/2, C] x = torch.cat([x0, x1, x2, x3], -1) # ————————> [B, H/2, W/2, 4*C] 在channael维度上进行拼接 x = x.view(B, -1, 4 * C) # [B, H/2*W/2, 4*C] x = self.norm(x) x = self.reduction(x) # [B, H/2*W/2, 2*C] return x"""MLP模块"""class Mlp(nn.Module): """ MLP as used in Vision Transformer, MLP-Mixer and related networks """ def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): super().__init__() out_features = out_features or in_features hidden_features = hidden_features or in_features self.fc1 = nn.Linear(in_features, hidden_features) self.act = act_layer() self.drop1 = nn.Dropout(drop) self.fc2 = nn.Linear(hidden_features, out_features) self.drop2 = nn.Dropout(drop) def forward(self, x): x = self.fc1(x) x = self.act(x) x = self.drop1(x) x = self.fc2(x) x = self.drop2(x) return x"""WindowAttention即为：SW-MSA或者W-MSA模块"""class WindowAttention(nn.Module): r""" Window based multi-head self attention (W-MSA) module with relative position bias. It supports both of shifted and non-shifted window. Args: dim (int): Number of input channels. window_size (tuple[int]): The height and width of the window. num_heads (int): Number of attention heads. qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0 proj_drop (float, optional): Dropout ratio of output. Default: 0.0 """ def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0.): super().__init__() self.dim = dim self.window_size = window_size # [Mh, Mw] self.num_heads = num_heads head_dim = dim // num_heads self.scale = head_dim ** -0.5 # define a parameter table of relative position bias # 创建偏置bias项矩阵 self.relative_position_bias_table = nn.Parameter( torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads)) # [2*Mh-1 * 2*Mw-1, nH] 其元素的个数===>>[(2*Mh-1) * (2*Mw-1)] # get pair-wise relative position index for each token inside the window coords_h = torch.arange(self.window_size[0]) # 如果此处的self.window_size[0]为2的话，则生成的coords_h为[0,1] coords_w = torch.arange(self.window_size[1]) # 同理得 coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # [2, Mh, Mw] coords_flatten = torch.flatten(coords, 1) # [2, Mh*Mw] # [2, Mh*Mw, 1] - [2, 1, Mh*Mw] relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # [2, Mh*Mw, Mh*Mw] relative_coords = relative_coords.permute(1, 2, 0).contiguous() # [Mh*Mw, Mh*Mw, 2] relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0 行标+（M-1） relative_coords[:, :, 1] += self.window_size[1] - 1 # 列表标+（M-1） relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 relative_position_index = relative_coords.sum(-1) # [Mh*Mw, Mh*Mw] self.register_buffer("relative_position_index", relative_position_index) # 将relative_position_index放入到模型的缓存当中 self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop) nn.init.trunc_normal_(self.relative_position_bias_table, std=.02) self.softmax = nn.Softmax(dim=-1) def forward(self, x, mask: Optional[torch.Tensor] = None): """ Args: x: input features with shape of (num_windows*B, Mh*Mw, C) mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None """ # [batch_size*num_windows, Mh*Mw, total_embed_dim] B_, N, C = x.shape # qkv(): -> [batch_size*num_windows, Mh*Mw, 3 * total_embed_dim] # reshape: -> [batch_size*num_windows, Mh*Mw, 3, num_heads, embed_dim_per_head] # permute: -> [3, batch_size*num_windows, num_heads, Mh*Mw, embed_dim_per_head] qkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) # [batch_size*num_windows, num_heads, Mh*Mw, embed_dim_per_head] q, k, v = qkv.unbind(0) # make torchscript happy (cannot use tensor as tuple) # transpose: -> [batch_size*num_windows, num_heads, embed_dim_per_head, Mh*Mw] # @: multiply -> [batch_size*num_windows, num_heads, Mh*Mw, Mh*Mw] q = q * self.scale attn = (q @ k.transpose(-2, -1)) # relative_position_bias_table.view: [Mh*Mw*Mh*Mw,nH] -> [Mh*Mw,Mh*Mw,nH] relative_position_bias = self.relative_position_bias_table[self.relative_position_index.view(-1)].view( self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1) relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # [nH, Mh*Mw, Mh*Mw] attn = attn + relative_position_bias.unsqueeze(0) # 进行mask，相同区域使用0表示；不同区域使用-100表示 if mask is not None: # mask: [nW, Mh*Mw, Mh*Mw] nW = mask.shape[0] # num_windows # attn.view: [batch_size, num_windows, num_heads, Mh*Mw, Mh*Mw] # mask.unsqueeze: [1, nW, 1, Mh*Mw, Mh*Mw] attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0) attn = attn.view(-1, self.num_heads, N, N) attn = self.softmax(attn) else: attn = self.softmax(attn) attn = self.attn_drop(attn) # @: multiply -> [batch_size*num_windows, num_heads, Mh*Mw, embed_dim_per_head] # transpose: -> [batch_size*num_windows, Mh*Mw, num_heads, embed_dim_per_head] # reshape: -> [batch_size*num_windows, Mh*Mw, total_embed_dim] x = (attn @ v).transpose(1, 2).reshape(B_, N, C) x = self.proj(x) x = self.proj_drop(x) return x""" SwinTransformerBlock"""class SwinTransformerBlock(nn.Module): r""" Swin Transformer Block. Args: dim (int): Number of input channels. num_heads (int): Number of attention heads. window_size (int): Window size. shift_size (int): Shift size for SW-MSA. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True drop (float, optional): Dropout rate. Default: 0.0 attn_drop (float, optional): Attention dropout rate. Default: 0.0 drop_path (float, optional): Stochastic depth rate. Default: 0.0 act_layer (nn.Module, optional): Activation layer. Default: nn.GELU norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, dim, num_heads, window_size=7, shift_size=0, mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm): super().__init__() self.dim = dim self.num_heads = num_heads self.window_size = window_size self.shift_size = shift_size self.mlp_ratio = mlp_ratio assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" self.norm1 = norm_layer(dim) # 先经过层归一化处理 # WindowAttention即为：SW-MSA或者W-MSA模块 self.attn = WindowAttention( dim, window_size=(self.window_size, self.window_size), num_heads=num_heads, qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop) self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) def forward(self, x, attn_mask): H, W = self.H, self.W B, L, C = x.shape assert L == H * W, "input feature has wrong size" shortcut = x x = self.norm1(x) x = x.view(B, H, W, C) # pad feature maps to multiples of window size # 把feature map给pad到window size的整数倍 pad_l = pad_t = 0 pad_r = (self.window_size - W % self.window_size) % self.window_size pad_b = (self.window_size - H % self.window_size) % self.window_size x = F.pad(x, (0, 0, pad_l, pad_r, pad_t, pad_b)) _, Hp, Wp, _ = x.shape # cyclic shift # 判断是进行SW-MSA或者是W-MSA模块 if self.shift_size > 0: # https://blog.csdn.net/ooooocj/article/details/126046858?ops_request_misc=&request_id=&biz_id=102&utm_term=torch.roll()%E7%94%A8%E6%B3%95&utm_medium=distribute.pc_search_result.none-task-blog-2~all~sobaiduweb~default-0-126046858.142^v73^control,201^v4^add_ask,239^v1^control&spm=1018.2226.3001.4187 shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) #进行数据移动操作 else: shifted_x = x attn_mask = None # partition windows # 将窗口按照window_size的大小进行划分，得到一个个窗口 x_windows = window_partition(shifted_x, self.window_size) # [nW*B, Mh, Mw, C] # 将数据进行展平操作 x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # [nW*B, Mh*Mw, C] # W-MSA/SW-MSA """ # 进行多头自注意力机制操作 """ attn_windows = self.attn(x_windows, mask=attn_mask) # [nW*B, Mh*Mw, C] # merge windows attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) # [nW*B, Mh, Mw, C] # 将多窗口拼接回大的featureMap shifted_x = window_reverse(attn_windows, self.window_size, Hp, Wp) # [B, H', W', C] # reverse cyclic shift # 将移位的数据进行还原 if self.shift_size > 0: x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) else: x = shifted_x # 如果进行了padding操作，需要移出掉相应的pad if pad_r > 0 or pad_b > 0: # 把前面pad的数据移除掉 x = x[:, :H, :W, :].contiguous() x = x.view(B, H * W, C) # FFN x = shortcut + self.drop_path(x) x = x + self.drop_path(self.mlp(self.norm2(x))) return xclass BasicLayer(nn.Module): """ A basic Swin Transformer layer for one stage. Args: dim (int): Number of input channels. depth (int): Number of blocks. num_heads (int): Number of attention heads. window_size (int): Local window size. mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True drop (float, optional): Dropout rate. Default: 0.0 attn_drop (float, optional): Attention dropout rate. Default: 0.0 drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. """ def __init__(self, dim, depth, num_heads, window_size, mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False): super().__init__() self.dim = dim self.depth = depth self.window_size = window_size self.use_checkpoint = use_checkpoint self.shift_size = window_size // 2 # 表示向右和向下偏移的窗口大小 即窗口大小除以2，然后向下取整 # build blocks self.blocks = nn.ModuleList([ SwinTransformerBlock( dim=dim, num_heads=num_heads, window_size=window_size, shift_size=0 if (i % 2 == 0) else self.shift_size, # 通过判断shift_size是否等于0，来决定是使用W-MSA与SW-MSA mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, drop=drop, attn_drop=attn_drop, drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path, norm_layer=norm_layer) for i in range(depth)]) # patch merging layer 即：PatchMerging类 if downsample is not None: self.downsample = downsample(dim=dim, norm_layer=norm_layer) else: self.downsample = None def create_mask(self, x, H, W): # calculate attention mask for SW-MSA # 保证Hp和Wp是window_size的整数倍 Hp = int(np.ceil(H / self.window_size)) * self.window_size Wp = int(np.ceil(W / self.window_size)) * self.window_size # 拥有和feature map一样的通道排列顺序，方便后续window_partition img_mask = torch.zeros((1, Hp, Wp, 1), device=x.device) # [1, Hp, Wp, 1] h_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) w_slices = (slice(0, -self.window_size), slice(-self.window_size, -self.shift_size), slice(-self.shift_size, None)) cnt = 0 for h in h_slices: for w in w_slices: img_mask[:, h, w, :] = cnt cnt += 1 # 将img_mask划分成一个一个窗口 mask_windows = window_partition(img_mask, self.window_size) # [nW, Mh, Mw, 1] # 输出的是按照指定的window_size划分成一个一个窗口的数据 mask_windows = mask_windows.view(-1, self.window_size * self.window_size) # [nW, Mh*Mw] attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) # [nW, 1, Mh*Mw] - [nW, Mh*Mw, 1] 使用了广播机制 # [nW, Mh*Mw, Mh*Mw] # 因为需要求得的是自身注意力机制，所以，所以相同的区域使用0表示，；不同的区域不等于0，填入-100，这样，在求得 attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) # 即对于不等于0的位置，赋值为-100；否则为0 return attn_mask def forward(self, x, H, W): attn_mask = self.create_mask(x, H, W) # [nW, Mh*Mw, Mh*Mw] # 制作mask蒙版 for blk in self.blocks: blk.H, blk.W = H, W if not torch.jit.is_scripting() and self.use_checkpoint: x = checkpoint.checkpoint(blk, x, attn_mask) else: x = blk(x, attn_mask) if self.downsample is not None: x = self.downsample(x, H, W) H, W = (H + 1) // 2, (W + 1) // 2 return x, H, Wclass SwinTransformer(nn.Module): r""" Swin Transformer A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - https://arxiv.org/pdf/2103.14030 Args: patch_size (int | tuple(int)): Patch size. Default: 4 表示通过Patch Partition层后，下采样几倍 in_chans (int): Number of input image channels. Default: 3 num_classes (int): Number of classes for classification head. Default: 1000 embed_dim (int): Patch embedding dimension. Default: 96 depths (tuple(int)): Depth of each Swin Transformer layer. num_heads (tuple(int)): Number of attention heads in different layers. window_size (int): Window size. Default: 7 mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4 qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True drop_rate (float): Dropout rate. Default: 0 attn_drop_rate (float): Attention dropout rate. Default: 0 drop_path_rate (float): Stochastic depth rate. Default: 0.1 norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm. patch_norm (bool): If True, add normalization after patch embedding. Default: True use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False """ def __init__(self, patch_size=4, # 表示通过Patch Partition层后，下采样几倍 in_chans=3, # 输入图像通道 num_classes=1000, # 类别数 embed_dim=96, # Patch partition层后的LinearEmbedding层映射后的维度，之后的几层都是该数的整数倍 分别是 C、2C、4C、8C depths=(2, 2, 6, 2), # 表示每一个Stage模块内，Swin Transformer Block重复的次数 num_heads=(3, 6, 12, 24), # 表示每一个Stage模块内，Swin Transformer Block中采用的Multi-Head self-Attention的head的个数 window_size=7, # 表示W-MSA与SW-MSA所采用的window的大小 mlp_ratio=4., # 表示MLP模块中，第一个全连接层增大的倍数 qkv_bias=True, drop_rate=0., # 对应的PatchEmbed层后面的 attn_drop_rate=0., # 对应于Multi-Head self-Attention模块中对应的dropRate drop_path_rate=0.1, # 对应于每一个Swin-Transformer模块中采用的DropRate 其是慢慢的递增的，从0增长到drop_path_rate norm_layer=nn.LayerNorm, patch_norm=True, use_checkpoint=False, **kwargs): super().__init__() self.num_classes = num_classes self.num_layers = len(depths) # depths:表示重复的Swin Transoformer Block模块的次数 表示每一个Stage模块内，Swin Transformer Block重复的次数 self.embed_dim = embed_dim self.patch_norm = patch_norm # stage4输出特征矩阵的channels self.num_features = int(embed_dim * 2 ** (self.num_layers - 1)) self.mlp_ratio = mlp_ratio # split image into non-overlapping patches 即将图片划分成一个个没有重叠的patch self.patch_embed = PatchEmbed( patch_size=patch_size, in_c=in_chans, embed_dim=embed_dim, norm_layer=norm_layer if self.patch_norm else None) self.pos_drop = nn.Dropout(p=drop_rate) # PatchEmbed层后面的Dropout层 # stochastic depth dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule # build layers self.layers = nn.ModuleList() for i_layer in range(self.num_layers): # 注意这里构建的stage和论文图中有些差异 # 这里的stage不包含该stage的patch_merging层，包含的是下个stage的 layers = BasicLayer(dim=int(embed_dim * 2 ** i_layer), # 传入特征矩阵的维度，即channel方向的深度 depth=depths[i_layer], # 表示当前stage中需要堆叠的多少Swin Transformer Block num_heads=num_heads[i_layer], # 表示每一个Stage模块内，Swin Transformer Block中采用的Multi-Head self-Attention的head的个数 window_size=window_size, # 表示W-MSA与SW-MSA所采用的window的大小 mlp_ratio=self.mlp_ratio, # 表示MLP模块中，第一个全连接层增大的倍数 qkv_bias=qkv_bias, drop=drop_rate, # 对应的PatchEmbed层后面的 attn_drop=attn_drop_rate, # 对应于Multi-Head self-Attention模块中对应的dropRate drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # 对应于每一个Swin-Transformer模块中采用的DropRate 其是慢慢的递增的，从0增长到drop_path_rate norm_layer=norm_layer, downsample=PatchMerging if (i_layer < self.num_layers - 1) else None, # 判断是否是第四个，因为第四个Stage是没有PatchMerging层的 use_checkpoint=use_checkpoint) self.layers.append(layers) self.norm = norm_layer(self.num_features) self.avgpool = nn.AdaptiveAvgPool1d(1) # 自适应的全局平均池化 self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity() self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): nn.init.trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def forward(self, x): # x: [B, L, C] x, H, W = self.patch_embed(x) # 对图像下采样4倍 x = self.pos_drop(x) # 依次传入各个stage中 for layer in self.layers: x, H, W = layer(x, H, W) x = self.norm(x) # [B, L, C] x = self.avgpool(x.transpose(1, 2)) # [B, C, 1] x = torch.flatten(x, 1) x = self.head(x) # 经过全连接层，得到输出 return xdef swin_tiny_patch4_window7_224(num_classes: int = 1000, **kwargs): # trained ImageNet-1K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=7, embed_dim=96, depths=(2, 2, 6, 2), num_heads=(3, 6, 12, 24), num_classes=num_classes, **kwargs) return modeldef swin_small_patch4_window7_224(num_classes: int = 1000, **kwargs): # trained ImageNet-1K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_small_patch4_window7_224.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=7, embed_dim=96, depths=(2, 2, 18, 2), num_heads=(3, 6, 12, 24), num_classes=num_classes, **kwargs) return modeldef swin_base_patch4_window7_224(num_classes: int = 1000, **kwargs): # trained ImageNet-1K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=7, embed_dim=128, depths=(2, 2, 18, 2), num_heads=(4, 8, 16, 32), num_classes=num_classes, **kwargs) return modeldef swin_base_patch4_window12_384(num_classes: int = 1000, **kwargs): # trained ImageNet-1K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=12, embed_dim=128, depths=(2, 2, 18, 2), num_heads=(4, 8, 16, 32), num_classes=num_classes, **kwargs) return modeldef swin_base_patch4_window7_224_in22k(num_classes: int = 21841, **kwargs): # trained ImageNet-22K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22k.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=7, embed_dim=128, depths=(2, 2, 18, 2), num_heads=(4, 8, 16, 32), num_classes=num_classes, **kwargs) return modeldef swin_base_patch4_window12_384_in22k(num_classes: int = 21841, **kwargs): # trained ImageNet-22K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384_22k.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=12, embed_dim=128, depths=(2, 2, 18, 2), num_heads=(4, 8, 16, 32), num_classes=num_classes, **kwargs) return modeldef swin_large_patch4_window7_224_in22k(num_classes: int = 21841, **kwargs): # trained ImageNet-22K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window7_224_22k.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=7, embed_dim=192, depths=(2, 2, 18, 2), num_heads=(6, 12, 24, 48), num_classes=num_classes, **kwargs) return modeldef swin_large_patch4_window12_384_in22k(num_classes: int = 21841, **kwargs): # trained ImageNet-22K # https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window12_384_22k.pth model = SwinTransformer(in_chans=3, patch_size=4, window_size=12, embed_dim=192, depths=(2, 2, 18, 2), num_heads=(6, 12, 24, 48), num_classes=num_classes, **kwargs) return model本文链接：https://www.kjpai.cn/gushi/2024-04-01/151967.html，文章来源：网络cs，作者：璐璐，版权归作者所有，如需转载请注明来源和作者，否则将追究法律责任！
版权声明：本文内容由互联网用户自发贡献，该文观点仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。